Dataset for: Video-Based Joint Dynamics and Machine Learning for Automated Gymnastics Score Estimation

This repository contains the processed datasets and analysis code associated with the study “Video-Based Joint Dynamics and Machine Learning for Automated Gymnastics Score Estimation.” The dataset includes extracted human pose keypoints obtained from video recordings using a computer vision–based pose estimation framework. Biomechanically meaningful joint-level features were computed, including joint displacement magnitude, variability measures, and range-of-motion descriptors for selected key joints. The repository also provides execution score labels assigned by certified judges, as well as scripts for feature extraction, preprocessing, model training, and evaluation. Raw video recordings are not publicly shared due to privacy and ethical considerations. The released data consist exclusively of processed pose representations and derived features that enable methodological transparency and reproducibility without exposing identifiable participant information. This dataset is intended to facilitate reproducible research in sports biomechanics, human motion analysis, and machine learning–driven performance evaluation in artistic gymnastics.